evolution and challenges of software defined … · evolution and challenges of software defined...
TRANSCRIPT
Evolution and Challenges of Software Defined Networking
EVOLUTION AND CHALLENGES OF SOFTWARE DEFINED NETWORKING
Ángel Leonardo Valdivieso
Lorena Isabel Barona
{angevald,lorebaro}@ucm.es
Group of Analysis, Security and Systems
Department of Software Engineering and Artificial Intelligence
Complutense University of Madrid
Evolution and Challenges of Software Defined Networking
Outline
1. Introduction
2. Evolution of Software Defined Networking
3. SDN Research Initiatives
4. SDN Challenges
5. Conclusions
2
Evolution and Challenges of Software Defined Networking
Introduction
3
Global Internet IP Traffic by Device Type
3. SDN Research
Initiatives
1. Introduction
5. Conclusions
4. SDN Challenges
2. Evolution of
SDN
Evolution and Challenges of Software Defined Networking
Introduction
4
Global Consumer IP Traffic
3. SDN Research
Initiatives
1. Introduction
5. Conclusions
4. SDN Challenges
2. Evolution of
SDN
Evolution and Challenges of Software Defined Networking
Introduction
5
The networks are ready to support this new requeriments?
New requirements:
Changing traffic patterns dynamically
Support to public and private cloud services
Intelligent network management of increasing bandwidth – Big data support
Easy support to implement custom network services 3. SDN Research
Initiatives
1. Introduction
5. Conclusions
4. SDN Challenges
2. Evolution of
SDN
Evolution and Challenges of Software Defined Networking
Introduction
6
Limitations of Current Networking Technologies:
Need to configure network devices individually (using CLI)
Vendor dependence to support new capabilities and services
Static nature of networks in contrast to the today´s dynamic environment (server virtualization, distributed applications)
Protocols solve a specific problem (ACL, VLAN, QoS, Firewall, NIDS, authentication…) complexity
DATA PLANE
CONTROL PLANE
DATA PLANE
CONTROL PLANE 3. SDN Research
Initiatives
1. Introduction
5. Conclusions
4. SDN Challenges
2. Evolution of
SDN
Evolution and Challenges of Software Defined Networking
Evolution of SDN
7
Active Networks:
Separation Control-Data:
Openflow and NOS:
(1990 – 2000) (2000-2010) (2010)
• Smart Packets
• SwitchWare
• Calvert…
• Tempest
• ForCES
• RCP
• Ethane…
• Openflow Protocol
• NOX, Beacon
• Procera…
• Open the control network through a programming interface (network API)
• Programming models: Node Operating System (NodeOS)
• Open interface between control and data planes
• Logically centralized control of the network
• Distributed state management
• Interface build on already supported technology (flowtables)
• Vision of a Network Operating System (NOS)
• Not offer practical performance and security
• Difficult implementation in production networks (compatible hardware)
• Difficult to enable sophisticated middlebox functionality
3. SDN Research
Initiatives
1. Introduction
5. Conclusions
4. SDN Challenges
2. Evolution of
SDN
Evolution and Challenges of Software Defined Networking
Evolution of SDN
8
Software Defined Networking is an emerging network architecture where network control is decoupled from forwarding and is directly programmable.
3. SDN Research
Initiatives
1. Introduction
5. Conclusions
4. SDN Challenges
2. Evolution of
SDN
Evolution and Challenges of Software Defined Networking
Evolution of SDN
9
Software Defined Networking is an emerging network architecture where network control is decoupled from forwarding and is directly programmable.
3. SDN Research
Initiatives
1. Introduction
5. Conclusions
4. SDN Challenges
2. Evolution of
SDN
Source: Open Networking Foundation (ONF)
Evolution and Challenges of Software Defined Networking
SDN Research Initiatives
10
Virtualization & NOS
Multimedia
Simulation
3. SDN Research
Initiatives
1. Introduction
5. Conclusions
4. SDN Challenges
2. Evolution of
SDN
Testbeds
Evolution and Challenges of Software Defined Networking
SDN Research Initiatives
11
Virtualization
3. SDN Research
Initiatives
1. Introduction
5. Conclusions
4. SDN Challenges
2. Evolution of
SDN
Evolution and Challenges of Software Defined Networking
SDN Research Initiatives
12
Virtualization: FlowVisor
OpenFlow Switch
FlowVisor Slicing Policy
OpenFlow Switch OpenFlow Switch
OpenFlow Protocol
NOX … Beacon
PROCERA FRENETIC …
APP APP …
Flowtable Flowtable Flowtable
FlowVisor Slicing Policy
DATA PLANE
3. SDN Research
Initiatives
1. Introduction
5. Conclusions
4. SDN Challenges
2. Evolution of
SDN
Evolution and Challenges of Software Defined Networking
SDN Research Initiatives
13
Network Operating System NOS
Switch, Virtual Switch, Router
Procera
3. SDN Research
Initiatives
1. Introduction
5. Conclusions
4. SDN Challenges
2. Evolution of
SDN
Evolution and Challenges of Software Defined Networking
SDN Research Initiatives
14
Simulation: Mininet - An Instant Virtual Network on your Laptop
3. SDN Research
Initiatives
1. Introduction
5. Conclusions
4. SDN Challenges
2. Evolution of
SDN
Source: B. Lantz, B. Heller, and N. McKeown, “A Network in a Laptop: Rapid Prototyping for Software-Defined Networks”
Evolution and Challenges of Software Defined Networking
SDN Research Initiatives
15
Testbeds: Global Environment for Network Innovation (GENI)
3. SDN Research
Initiatives
1. Introduction
5. Conclusions
4. SDN Challenges
2. Evolution of
SDN
Evolution and Challenges of Software Defined Networking
SDN Research Initiatives
16
Testbeds: OpenFlow in Europe
Linking Infrastructure and Applications (OFELIA) 3. SDN Research
Initiatives
1. Introduction
5. Conclusions
4. SDN Challenges
2. Evolution of
SDN
Evolution and Challenges of Software Defined Networking
SDN Research Initiatives
17
Multimedia: SDN helps to improve the QoE experience
SDN Path Optimization Systems Architecture: QoS Matching and Oprtimization Function (QMOF) and Path Assignment Function (PAF)
If the quality of the network is degraded, the system can react and dynamically modify the path parameters of the network prioritizing the users configuration
3. SDN Research
Initiatives
1. Introduction
5. Conclusions
4. SDN Challenges
2. Evolution of
SDN
Source: A. Kassler, L. Skorin-Kapov, O. Dobrijevic, “Towards QoE-driven Multimedia Service Negotiation and Path Optimization with Software Defined Networking,”
Evolution and Challenges of Software Defined Networking
SDN Research Initiatives
18
Multimedia: Fast Recovery of a link failure
SDN improves the recovery time from a link failure using an alternative path
The system uses the restoration and protection mechanism.
In restoration system, the controller calculates and establishes an alternative path when a path failure signal is received.
In protection system, the controller anticipates a failure and calculates two disjoint paths (working and protected)
The experiments show a recovery time of less than 50 ms for carrier-grade network.
3. SDN Research
Initiatives
1. Introduction
5. Conclusions
4. SDN Challenges
2. Evolution of
SDN
Source: S. Sharma, D. Staessens, D. Colle, M. Pickavet, and P. Demeester, “A Demonstration of Fast Failure Recovery in Software Defined Networking,”
Evolution and Challenges of Software Defined Networking 19
How many controllers needed and the localization of these controllers?
SDN Challenges
Modeling and Performance:
SDN establishes the separation of control and data planes, but doesn’t necessarily order the existence of a single controller.
Control plane may have a single controller, a set of controllers of even any dynamic topology
The number and position of the controllers depends on desired reaction bounds, metric choice and the network topology
3. SDN Research
Initiatives
1. Introduction
5. Conclusions
4. SDN
Challenges
2. Evolution of
SDN
Evolution and Challenges of Software Defined Networking 20
How to select the appropriate NOS and how to measure its performance?
SDN Challenges
Modeling and Performance:
It is necessary to establish a balance between high performance and the productivity of developers (a developer friendly language)
The performance of the controller depends directly of programming language (C++, Java, Python) and the development environment
3. SDN Research
Initiatives
1. Introduction
5. Conclusions
4. SDN
Challenges
2. Evolution of
SDN
Evolution and Challenges of Software Defined Networking 21
SDN Challenges
NOS: How to measure its performance?
How many datapath request can the controller handle per second?
How fast the controller responds to datapath request?
3. SDN Research
Initiatives
1. Introduction
5. Conclusions
4. SDN
Challenges
2. Evolution of
SDN
Source: D. Erickson, “The Beacon Openflow Controller,”
Evolution and Challenges of Software Defined Networking 22
A failure of the controller can negatively compromise resilience of the whole network.
SDN Challenges
Resilience and Recovery:
The failure can be fired by abruptly aborting of a NOS process, an unexpected error of the application or a DDoS attack.
Openflow establishes the configuration of one or more backup controllers, but doesn’t provide a coordination mechanism between them.
The project CPRecovery permits the replication between primary and secondary SDN controller in two phases.
The replication phase acts during normal functioning of the controller and send regular updates to the backup controller. The recovery phase acts during a failure state of the primary controller and starts the backup controller as a main controller.
3. SDN Research
Initiatives
1. Introduction
5. Conclusions
4. SDN
Challenges
2. Evolution of
SDN
Source:P. Fonseca, R. Bennesby, E. Mota, and A. Passito, “A Replication Component for Resilient OpenFlow-based Networking,”
Evolution and Challenges of Software Defined Networking 23
Integration with legacy networks (actual Internet infrastructure)
SDN Challenges
Convergence and Integration:
The equipment could be upgraded or even replaced to support Openflow High costly network re-engineering
It is necessary a coordination between SDN equipment and legacy infrastructure
OpenFlow is unable to handle or manage legacy equipment it is difficult to connect two-Openflow environments with legacy networks.
3. SDN Research
Initiatives
1. Introduction
5. Conclusions
4. SDN
Challenges
2. Evolution of
SDN
Source:F. Farias, J. Salvatti, E. Cerqueira, “A Proposal Management of the Legacy Network Environment Using Openflow Control Plane,”
Evolution and Challenges of Software Defined Networking
Conclusions
24
SDN can be expanded beyond actual match/action paradigm. For example, it could integrate middleboxes or programmable custom packet processors.
This integration could offers new services like on the fly encryption, transcoding, or traffic classification. This require coordination, consensus and vendor support.
In control plane, the composing and coupling of heterogeneous components are still difficult. For example, compose application using Beacon, POX or Floodlight simultaneously.
Finally, remark that SDN is a tool. The research community can use this tool to create new innovative services and applications.
3. SDN Research
Initiatives
1. Introduction
5. Conclusions
4. SDN Challenges
2. Evolution of
SDN
Evolution and Challenges of Software Defined Networking 25
Group of Analysis, Security and Systems
Department of Software Engineering and Artificial Intelligence
Complutense University of Madrid
Ángel Leonardo Valdivieso [email protected]
Thanks for your time and attention !